PHYSICS REVIEW

Susceptibility-Weighted Imaging: TechnicalAspects and Clinical Applications, Part 1

E.M. HaackeS. Mittal

Z. WuJ. Neelavalli

Y.-C.N. Cheng

SUMMARY: Susceptibility-weighted imaging (SWI) is a new neuroimaging technique, which usestissue magnetic susceptibility differences to generate a unique contrast, different from that of spindensity, T1, T2, and T2*. In this review (the first of 2 parts), we present the technical background forSWI. We discuss the concept of gradient-echo images and how we can measure local changes insusceptibility. Armed with this material, we introduce the steps required to transform the originalmagnitude and phase images into SWI data. The use of SWI filtered phase as a means to visualize andpotentially quantify iron in the brain is presented. Advice for the correct interpretation of SWI data isdiscussed, and a set of recommended sequence parameters for different field strengths is given.

Susceptibility-weighted imaging (SWI) is a new means toenhance contrast in MR imaging.1-56 Until recently, with

the exception of phase being used for large-vessel flow quan-tification or for use in inversion recovery sequences, most di-agnostic MR imaging relied only on the reading of magnitudeinformation. The phase information was ignored and usuallydiscarded before even reaching the viewing console. Phase im-ages, however, contain a wealth of information about localsusceptibility changes between tissues,1,22,26,57-63 which can beuseful in measuring iron content26 and other substances thatchange the local field. The effects of other background mag-netic fields presented a major problem by obscuring the usefulphase information. Hence, for nearly 20 years, phase informa-tion in flow-compensated sequences went essentially unusedas a means to measure susceptibility in clinical MR imaging.

In 1997, we developed a means to remove most of the un-wanted phase artifacts and keep just the local phase of inter-est.1 This set the stage for discovering other aspects of suscep-tibility imaging and even hinted at the future direction ofsusceptibility mapping.26 We began to see different types ofcontrast in the SWI filtered phase images for different diseases.To make the information more accessible to radiologists,though phase and magnitude images separately are also criti-cal pieces of information, we combined the phase and themagnitude information and thus created a new susceptibility-weighted magnitude image, which today is referred to asSWI.22 This new pair of images (the SWI filtered phase dataand the merged SWI magnitude data) is now available on atleast 1 major manufacturer’s systems (Siemens Medical Sys-tems), though it took almost 10 years to see the actual imple-mentation of this approach in a clinical setting around theworld.

Many technical developments take years to leave the re-search environment and become part of standard radiologyapplications. This has been true for perfusion imaging, diffu-sion tensor imaging, magnetization transfer imaging, and alsofor SWI. The common feature of these methods is that they

originated as good scientific ideas; they were simulated at firstand then tested on phantoms, on volunteers, and finally on afew patients with the appropriate institutional review boardapprovals from the local institution where the research tookplace. These methods became scientifically adopted on a largerscale once they showed how clinically capable they were. Ifsuccessful, the National Institutes of Health would grant fundsfor broader testing of these methods. Finally, with proof of thevalue of the method in place and with a stable software pack-age, it would be possible for hundreds of institutions to beginusing these methods for diagnostic purposes. However, eventhen, the final goal will not have been achieved. The methodwould still need to be adopted in a broad clinical environmentand then executed as a standard “neuroprotocol” on a clinicalplatform. Once the latter step has taken place, the methodshould eventually be adopted by the radiology community as awhole and become part and parcel of conventional educa-tional courses offered at radiology meetings for accreditation.Only then could we say that a method is fully incorporatedinto the international radiology routine.

In this review, we summarize the motivation behind SWI, theconcepts that led to its discovery, the optimal parameters to use,and the interpretation of the SWI processed data and phase im-ages; and we conclude by discussing the technology in use today,the postprocessing, and future directions of SWI research.

MR imaging has already an overabundance of different ap-proaches for investigational anatomic, functional, and meta-bolic imaging. Each method focuses on a new issue, such as thedelivery of blood to the tissue in perfusion-weighted imagingor molecular motion along a fiber track in diffusion tensorimaging. The susceptibility information is an adjunct to whatis available with conventional spin density, T1-, and T2-weighted imaging methods. SWI offers information about anytissue that has a different susceptibility than its surroundingstructures such as deoxygenated blood, hemosiderin, ferritin,and calcium. There are numerous neurologic disorders thatcan benefit dramatically from a very sensitive method thatmonitors the amount of iron in the brain, whether in the formof deoxyhemoglobin, ferritin, or hemosiderin. Such diseasesand conditions include, but are not limited to, aging, multiplesclerosis (MS), stroke, trauma, vascular malformations, andtumors. There is no doubt that as SWI becomes more broadlyaccepted, we will see many new applications develop becauseof the astute observations of clinical researchers and the wideavailability of information for the whole spectrum of neuro-logic diseases seen on a daily basis in the clinical setting.

From the Departments of Radiology (E.M.H., Y.-C.N.C.), Biomedical Engineering (E.M.H.,J.N., Y.-C.N.C.), and Neurosurgery (S.M.), Wayne State University, Detroit, Mich; KarmanosCancer Institute (S.M.), Detroit, Mich; and Department of Biomedical Engineering (Z.W.),McMaster University, Hamilton, Ontario, Canada.

This work was supported in part by the National Institutes of Health grant NIH R01 NHLBI62983-04 and Siemens Medical Systems.

Please address correspondence to E. Mark Haacke, PhD, Wayne State University, Detroit,MI; e-mail: nmrimaging@aol.com

DOI 10.3174/ajnr.A1400

PHYSICS

REVIEW

AJNR Am J Neuroradiol 30:19 –30 � Jan 2009 � www.ajnr.org 1

Published November 27, 2008 as 10.3174/ajnr.A1400

Copyright 2008 by American Society of Neuroradiology.

Gradient-Echo ImagingHistorically, data were first collected with a free induction de-cay.64,65 The concept of a spin-echo was invented only when itwas seen that local field inhomogeneities had caused dramaticsignal intensity loss. For some time, the spin-echo approachwas the mainstay of clinical MR imaging for resistive ma-chines.66 Not long after the appearance of superconductingsystems on the market, gradient-echo imaging67-69 came backto life because the field uniformity in these systems was mark-edly better (and has continued to improve during the last 20years). People quickly realized that they could eliminate mostartifacts in magnitude images by running the gradient-echoscans with short TEs in sequences like magnetization-prepa-ration rapid acquisition of gradient echo.70 This high-band-width sequence had little distortion and offered excellent T1contrast (and presented a good method to collect T1-weighteddata even at 3T). Longer TEs were used in 2D gradient-echoimaging but with fairly thick sections, usually 3 or 4 mm. TheTEs were often chosen to be 15, 20, or rarely 25 ms becauseof the signal-intensity loss associated with rapid phasechanges across a voxel. Imaging with longer TEs, whilemaintaining image quality, was only possible with 3D gra-dient-echo imaging. Once this was recognized and resolu-tions on the order of �1 mm3 were realized with a sufficientsignal intensity–to-noise ratio (SNR), the ability to look athidden information in phase data became possible.57,71

Thus, the stage to study susceptibility effects was set in themid 1990s, and the first article on SWI was published in1997.1 To appreciate the basic elements of SWI, we mustfirst review the basic gradient-echo sequence design onwhich it is based.

In MR imaging, it is necessary to create some transversemagnetization first.72 This can be accomplished by using atransmit coil to produce a radio-frequency (rf) pulse whichtips the longitudinal magnetization (Mo) partially or com-pletely into the transverse plane. The rf pulse is applied at theLarmor frequency, � � �Bo. If Mo is rotated an angle � aboutthe x-axis, it will create a component along y that will be pickedup by a receiver coil. The data are then encoded in both thephase- and partition-encoding directions (y and z, respec-tively, for a transverse acquisition), while a dephasing gradientis applied in the readout direction. The read gradient is thenturned on and the data are sampled. After this, the remnantsignal intensity is further dephased, and rf spoiling is appliedby varying the axis about which the rf is applied so that notransverse magnetization remains before the next rf pulse. Allthis is shown pictorially in Fig 1. The magnitude signal-inten-sity response for this rf-spoiled short-TR gradient-echo se-quence is given by

1) �m(�) � �osin� exp(�TE/T2*)�

[1 � exp(�TR/T1)] / [1 � cos� exp(�TR/T1)]

where �o is the tissue spin density, TR is the repeat time of eachdata acquisition, T1 is the tissue longitudinal relaxation time,and � is the angle by which the magnetization is tipped (usu-ally called the flip angle). A signal-intensity behavior exampleis given for white matter, gray matter, and CSF for 1.5T and 3T(Fig 2 and Table 1). The full signal-intensity response is givenby (for a right-handed system)

2) ��� � � �m(�)exp(�i��BTE),

where �B represents the local field deviation caused by iron,for example, and � is the gyromagnetic constant, which isequal to 2� � 42.58 MHz/T for protons. This can be rewrittenas

3) ��� � � �m(�)exp(�i�g��BoTE),

where g is a geometric factor, �� is the difference of the localmagnetic susceptibility of the tissue of interest from its sur-roundings, and B0 is the field strength. Starting with the Lar-mor equation

4) � � �Bo,

the expression for phase is

5) � �t.

The change in phase between 2 tissues after a period of time,TE, can then be written as

� � ��TE,

where

�� � ��B and �B � g��Bo

so that

6) � � ��BTE

7) � � �g��BoTE.

All of �B, ��, and, therefore, � are dependent on the localtissue susceptibility.

Magnetic SusceptibilityMagnetic susceptibility is defined as the magnetic response ofa substance when it is placed in an external magnetic field.When an object is placed inside a uniform magnetic field, theinduced magnetization, M, is given by

8) M � �H,

where

Fig 1. Gradient-echo sequence design. Ts refers to the sampling time interval. The gradientpulses are designed to give first-order flow compensation. SS indicates the slice-selectgradient; PE, the phase-encoding gradient; ADC, analog-to-digital conversion.

2 Haacke � AJNR 30 � Jan 2009 � www.ajnr.org

9) H � B / �o � M

and � is the magnetic susceptibility relating M and H. We canalso write

10) B � �H,

where � � �o�r is the permeability, �o is the permeability invacuum, �r is the relative permeability, and H the magneticfield (B is the induced magnetic field). Using the relative per-meability as

11) �r � 1 � �,

we find

12) M � �B/�o/(1 � �).

For linear materials with � �� 1, we find

13) M � �B/�o,

and it becomes clear that the induced magnetization is directlyproportional to the main field and the magnetic susceptibility.Each tissue or substance behaves somewhat differently in amagnetic field. For the purposes of this review, we shall con-sider only diamagnetic and paramagnetic substances. Dia-magnetic substances have � � 0. For example, calcium in thebody tends to be in the form of calcium phosphates such asCa3(PO)4 and is diamagnetic. On the other hand, paramag-netic substances (ie, most iron-based materials or tissues) have� 0. For example, deoxygenated blood has �� 0.45 ppmin SI units, relative to surrounding tissue. Moreover, iron-containing proteins, like ferritin and hemosiderin, tend tohave varying susceptibilities depending on the amount of se-questered iron within the protein. Typically, on the basis ofphase measurements from 75 subjects,62 approximately a�� � 0.21 ppm in SI units with respect to CSF, corresponds to60 �g of iron (Fe) per gram of tissue (although this may un-

derestimate the amount of Fe by a factor of 5). The suscepti-bilities of most biologic tissues73 tend to be �10�4. However,all forms of iron may not be visible with MR imaging if �� �0 for that substance. This can happen if the iron is shielded byother elements such as oxygen. For a more detailed descrip-tion of the different forms of magnetism see Schenck73 andSaini et al.74

Geometry EffectsThe induced magnetization (owing to its susceptibility) in anobject within a uniform external magnetic field distorts theuniform field outside the object. The spatial distribution ofthis deviation in the external applied field is a function of thegeometry of the object.75 The local field deviation inside andaround an object is of interest because it gives rise to localphase differences in MR imaging. When discussing the effectsof geometry on local field variations, we usually neglect thebackground field and assume that � �� 1. For a cylinder mak-ing an angle � to the main field, the effect on the field inside thecylinder is given by

14) �Bin � ��Bo(3cos2� � 1)/6 � �eBo/3,

which includes the Lorentzian sphere term. We have used�� � �i � �e. Here, �i and �e are the susceptibilities inside andoutside the cylinder, respectively. Generally, when we arelooking at changes in local tissue fields, the final term �eBo / 3no longer plays a role because it is a common constant bothinside and outside the object. If we write

15) �Bin � gin��Bo,

we have

16) gin � (3cos2� � 1)/6.

For blood vessels in particular, we have

17) �Bin � ��doBo(3cos2� � 1)/6,

where

18) ��do � 4�A(0.18ppm)Hct(1 � Y).

Here, A is a constant to be determined by the absolute suscep-tibility of blood in vivo, Hct is the hematocrit, and Y is theoxygen saturation. Practically, the choice of A depends on thevalue of Y in vivo: with A � 1, then Y � 0.55,1 whereas withA � 1.5, then Y � 0.70.76,77

Fig 2. A, Plots showing the signal-intensity behavior as a function of flip angle for gray matter (GM) and white matter (WM) at field strengths 1.5T and 3T. B, Plots showing the correspondingcontrast between GM-WM and GM-CSF. These curves are calculated taking into consideration the higher signal intensity available at 3T and assuming that bandwidth is correspondinglyincreased at 3T to ensure equivalent geometric distortion. Tissue parameters are given in Table 1. Note that the GM/WM contrast is highest around 20° at 3T and there is a minimumor reversal of contrast around 5° (where GM is now brighter than WM; ie, it is a high-contrast spin echo-weighted image).

Table 1: Tissue parameters used to simulate the T1-weightedimages shown in Fig 2

GM WM CSFSpin density % 80 65 1001.5T 1000 ms 650 ms 4000 ms3T 1283 ms 838 ms 4000 ms

Note:—GM indicates gray matter; WM, white matter; CSF, cerebrospinal fluid. CSF istaken to have 100% water content.

AJNR Am J Neuroradiol 30:19 –30 � Jan 2009 � www.ajnr.org 3

The field outside is more complicated and is given by

19) �Bout � ��doBosin2�cos(2�)a2/(2r2),

where a is the radius of the blood vessel, which is assumed as along cylinder intercepting an angle � with the main field, r isthe distance perpendicularly from the axis of the cylinder tothe position, and � is the angle between the vector r and theprojection of the main field direction onto a plane perpendic-ular to the axis of the vessel. The constant term, �eBo / 3, hasbeen dropped in equation 19.

For more complicated structures, the expressions for mag-netic field variations around the object are not so straightfor-ward. The key point is that susceptibility differences between 2adjacent structures lead to a spatial field deviation within andaround them, which is a function of their geometries. Follow-ing equation 6, this field deviation in turn leads to local phasechanges in the MR phase images. However, phase images con-tain information about all magnetic fields, microscopic andmacroscopic. The former may be from small amounts of localiron deposits, for example, whereas the latter consists of fieldchanges caused by the geometry of the object, such as air/tissueinterface effects, and by inhomogeneities in the main magneticfield. With all these properties brought together, the phaseshould be written as

20) � ���Blocal geometry � �Bcs � �Bglobal geometry

� �Bmain field�TE.

All these terms are a function of the spatial location. The firstterm contains the local field changes of interest. The secondterm represents a chemical shift effect and causes changes ofthe form

21) �Bcs � � B0,

where represents the chemical shift. The difference betweenchemical shift and the first term in equation 20 is that chemicalshift has no geometry dependence but is more of a point fieldchange.75 The last 2 terms create artifacts and are not our in-terest. They tend to have a low spatial-frequency dependence(the phase varies slowly over the image) and, for this reason,can be removed with a high-pass (HP) filter.60

Creating Susceptibility-Weighted Filtered-Phase ImagesFirst, an HP filter is applied to remove the low-spatial fre-quency components of the background field. This is usuallydone by using a 64 � 64 low-pass filter divided into the orig-inal phase image (eg, an image with a matrix size of 512 � 512)to create an HP-filter effect. The end result is the HP filter.Specifically, the HP-filtered image is obtained by taking theoriginal image �(r), truncating it to the central n � n compleximage �n(r), creating an image by zero-filling the elementsoutside the central n � n elements, and then complex dividing�(r) by �n(r) to obtain a new image,60

22) ���r� � �(r)/�n(r),

which has most of the bothersome parts of the terms�Bglobal geometry � �Bmain field removed (Fig 3).

Once we have a pristine phase image with the backgroundfield changes removed, the door is open to differentiate onetype of tissue from another, depending on their susceptibili-ties. This may be particularly useful at high field because the T1difference between tissues begins to decrease, but any changesin susceptibility remain the same. Figure 4 shows an exampleof a T1-weighted image compared with the gradient-echomagnitude and phase data. Here we see that the gray matter,especially in the motor cortex area—the central sulcus—isclearly delineated in the phase image because of its iron con-tent. Figure 5 shows an example of a phase image in themidbrain.

Marrying Magnitude and Phase Images to CreateMagnitude SWI DataAlthough HP-filtered phase images themselves are quite illus-trative of the susceptibility contrast, they are influenced byboth the intratissue phase (from�Bin) and the associated extra-tissue phase (from �Bout), which can impair the differentia-tion of adjacent tissues with different susceptibilities. Hence,the “phase mask” is designed to enhance the contrast in theoriginal magnitude image by suppressing pixels having certainphase values. Magnitude and phase data are brought togetheras a final magnitude SWI dataset by multiplying a phase maskimage into the original magnitude image.22

The phase mask is designed to be a number between zeroand 1. When there are no phase characteristics that we want to

Fig 3. A, Raw phase image. B, HP-filtered phase image with a central filter size of 32 � 32; C, Filtered-phase image with a central filter size of 64 � 64.

4 Haacke � AJNR 30 � Jan 2009 � www.ajnr.org

enhance, the phase mask is set to 1. Otherwise, the phase maskis designed to suppress the signal intensity in areas where thephase takes on certain values. For example, if the phase ofinterest is negative, the mask is designed as

23) f(x) � [� � (x)]/� for �� � (x) � 0

� 1 otherwise,

where (x) is the phase at location x. The phase mask can beapplied any number of times to the original magnitude image�(x) to create a new image,

24) � � x� � fm(x)�(x)

with a different contrast. The number of phase mask multipli-cations is chosen to optimize the contrast-to-noise ratio of theSWI image. This mathematical masking process is illustratedin Fig 6 and with a specific example from a 1.5T dataset shownin Fig 7.

A simple mathematical model can be built to estimate thevalue of m (from Equation 24).30 The number of multiplica-tions needed for a phase �0.3� is �4. For a phase �0.3�, avalue of m � 4 should be used. However, choosing m � 4 for

a phase of 0.1� or � causes �25% reduction in visibility ineither case, suggesting that m � 4 is a good compromise if oneis limited to just 1 representative SWI dataset. Of course, mcan be varied to produce a series of SWI data that can be usedin a diagnosis. After this processing, the data are often viewedby using a minimum intensity projection (mIP) over �4 im-ages. Clinically, 4 sections are the current common display,though showing a projection over 10 sections can be useful tovisualize a larger coverage of the venous drainage system in agiven area. Figure 8 shows projection data over the originalmagnitude images and the SWI processed images. One canalso present the data as a sliding mIP, which means that onestarts with the original series of images and mIPs over 4, movesto the next original image and mIP over 4 again so that a seriesof 64 images now becomes a series of 61 mIP images.

Furthermore, combining phase and magnitude informa-tion ensures the detectability of signal-intensity changes com-ing from both T2 and susceptibility differences between tis-sues. The existence of a phase difference does not necessarilylead to a T2* effect (uniform phase causes no T2� decay), sothese 2 sources of information complement each other. Also,suppressing the veins fully helps to create a clear distinction

Fig 4. Short-echo T1-weighted image (A), compared with the SWI long-echo gradient-echo processed magnitude (B ) and HP-filtered phase data (C ).

Fig 5. A, HP-filtered phase image in the midbrain acquired at 4T with TE � 15 ms. B, India ink-stained cadaver brain section showing a strong correspondence to variations in signal intensityas seen with SWI HP-filtered phase.

AJNR Am J Neuroradiol 30:19 –30 � Jan 2009 � www.ajnr.org 5

between veins and arteries (Fig 7D�F). Therefore, we see thatboth magnitude and phase information are essential forproper tissue characterization, hence their marriage to createan SWI processed image.

Advanced Phase ProcessingThe HP-filtering method described earlier is quite successfulin removing the spatially slowly varying phase effects that ob-scure the local intertissue phase differences of interest. How-ever, apart from removing the spatially slowly varying back-ground field effects, the HP filter (depending on filter size)also removes some of the physiologically or pathophysiologi-cally relevant phase information from larger anatomic struc-tures. For this reason, it is rare to use a central filter larger than64 � 64 if one is interested in measuring iron content. If themain goal is just to create better MR venograms, then it isreasonable to use a central filter of 96 � 96 or 128 � 128without much adverse effect on the small vessels.

The larger the filter size, the more background field effectscan be removed, but also the concomitant loss of phase infor-mation from larger tissue structures is more probable. Ideally,background field effects could be removed by predicting thephase expected for a given individual’s global head geometry.This would leave only small low-spatial-frequency phase er-rors, allowing a smaller filter size to be used to create the finalHP result.78 An example of this novel approach is shown in Fig9. Although this currently is a time-consuming process, itwould not be surprising to see this method implemented in a

practical fashion in the coming years, making SWI more ro-bust to air/tissue interface field effects. In that case, it may bepossible to tackle problems such as subarachnoid hemor-rhages, subdural hemorrhages, and possibly spine imaging.

Another important consideration is measuring iron. Cur-rently, phase, measured from filtered MR images, is found tobe the most sensitive tool for measuring relative changes iniron content in the subcortical structures of the brain.62 Mea-suring iron content depends on the assignment of a givenphase change in a tissue to a known amount of iron in thattissue. In Haacke et al,62 this was done by associating the phasechange in the motor cortex with 60 �g of nonheme Fe/g tissue.This ignores the fact that some of the phase shift, if not most ofit, may come from heme iron and not ferritin. If this is indeedthe case, the same phase shift seen in globus pallidus minusany heme iron effect may have a slope 5 times smaller. In anycase, the reader needs to be aware that the absolute relation-ship between changes in phase and iron has not yet been es-tablished and that the relative contributions between hemeand nonheme iron have not been sorted out yet. On the otherhand, if blood levels remain constant and ferritin concentra-tion does change, then any relative values of iron change inpercentages would still be accurate, independent of whetherthis scale factor of 5 is present or not.

One problem in trying to compare phase information from1 person to another is the geometry dependence of the phasecoming from the shape of local objects of interest. If one as-sumes that the geometries of the structures of interest areroughly the same from person to person, then this might notbe a major concern, but it is a potential problem when tryingto measure absolute iron content. (It is less likely to be a prob-lem when measuring relative changes in iron or calcium con-tent.) The susceptibility of a structure is independent of theorientation, size, and shape of the structure; hence, a means toquantify tissue susceptibility would be of utmost importancebecause it will be a reliable measure of iron content. Tissuesusceptibility will be an important means to quantify iron andis a topic of considerable interest in the field today.26,27

Recommended Imaging Parameters at Different FieldStrengthsSWI can be used best at higher field strengths for a number ofreasons. First, at low fields, TEs need to be much longer. Forexample, to obtain optimum contrast for veins at 1.5T, oneneeds a TE of between 40 and 80 ms. If we try to keep theproduct ��BoTE constant so that the phase effect remains thesame from 1 field strength to another, then the minimum TEat 3T becomes only 20 ms. This can be expressed mathemati-cally as

25) �B0) � ��g��B0TE � constant,

which is tantamount to forcing

26) B0TE � B0�TE�,

where B0TE and B0�TE� refer to the products of field strengthand TE at 1.5T and 3T, respectively. This shorter TE has sev-eral advantages. First, the scanning at 3T can be run twice as fastcompared with 1.5T such that in the same period of time, twicethe number of sections can be obtained at 3T or an isotropicin-plane resolution can be obtained. Second, at high fields, the

Fig 6. Pictorial depiction of the phase-masking process. A, Phase profile in a filtered-phaseimage. B, Corresponding profile of the mask created from A. Once the mask is raised to thefourth power, the vein that has a phase of �� / 2 and a mask value of 0.5 will become1-/16 and, therefore, this vein will be almost as well suppressed as the vein with a phaseof �� and f(x) � 0. GM indicates gray matter; WM, white matter.

6 Haacke � AJNR 30 � Jan 2009 � www.ajnr.org

SNR increases, so a sacrifice of some SNR (because the imagequality at 1.5T is already good enough clinically) to obtain notonly faster coverage but also higher resolution of the whole braincoverage at 3T can be easily tolerated. With parallel imaging at 3T,whole-brain coverage by using SWI can be obtained in as short as4 minutes. Figure 10 shows the phase behavior as a function offield strength, given that equation 25 is obeyed. As expected, it isdifficult to differentiate field strengths except for the fact that theSNR is higher for the 3T image than the 1.5T image. Otherwise,the phase images should have the same contrast.

The advent of very high fields such as 7T and above offersunprecedented resolution and SNR in imaging the brain.However, there are concerns about field homogeneity andhigh specific absorption rates (SAR). Gradient-echo imaging

is one of the most robust sequences in MR imaging and con-tinues to show its advantages at high field where images appearmuch more uniform than spin-echo images. SWI, for exam-ple, uses flip angles much smaller than 90°, and because theSAR varies as the square of the flip angles, SWI has a muchsmaller power deposition than spin-echo sequences. As dis-cussed previously, phase can be kept invariant as a function offield strength by appropriately adjusting the TE. On the otherhand, phase information from local structures does notstrongly depend on local rf inhomogeneities. These 2 proper-ties suggest that the SWI filtered-phase images can play a rolein comparing subject data across field strengths. High-fieldimages are shown in Figs 11 and 12 (for recommended imag-ing parameters at different field strengths, see Table 2). From

Fig 7. A, Unprocessed original SWI magnitude image. B, HP-filtered phase image. C, Processed SWI magnitude image (ie, after phase-mask multiplication with m � 4). D, Unprocessedoriginal SWI magnitude image showing the dark hypointense ring around vein cross-sections (arrows). E, HP-filtered phase image showing the veins have low signal-intensity (arrows).F, Processed SWI magnitude image showing that the veins now appear with uniform low signal intensity (arrows).

Fig 8. mIP data over the original magnitude images (A) and over the processed SWI images (B ). The mIP is carried out over 7 sections (representing a 14-mm slab thickness). The imageswere collected at 3T. There is a dropout of signal intensity in the frontal part of the brain, but otherwise the vessel visibility is much improved in B. In the future, this frontal dropoutshould not be a problem when the air/tissue geometries are corrected (Fig 9).

AJNR Am J Neuroradiol 30:19 –30 � Jan 2009 � www.ajnr.org 7

Fig 9. Results of the improved processing methodology. A, Original SWI phase image. B, Simulated phase due to air/tissue geometry at 40 ms. C, Result of subtracting A from B throughthe complex division. D, Result of HP filtering of A. E, Result of HP filtering of C. F, Corresponding unprocessed magnitude SWI image. G, Processed SWI magnitude image by using aphase mask from D. H, Processed SWI magnitude by using a phase mask from E. I, Result of subtracting G from H. Corresponding magnitude and phase images have been adjusted tothe same, but appropriate, contrast levels. The size of the central filter in the HP process is 64 � 64.

Fig 10. A and B, Phase images, keeping the product BoTE constant. A phase image acquired at 1.5T (A) and the same subject at 3T (B ) show the same overall contrast but with a betterSNR. The gray/white matter contrast in these images comes from the increased MR-visible iron content in the gray matter, giving it an appearance similar to a T1-weighted scan.

8 Haacke � AJNR 30 � Jan 2009 � www.ajnr.org

the perspective of higher SNR, smaller changes in phase nowbecome visible, and we are beginning to discover new phasevariations that reveal venules and arcuate fibers similar towhat one sees in stained cadaver brain studies.

Interpreting SWI DataThere are 3 components to interpret in the SWI data to makea clinical diagnosis. The first relates to the original magnitudeimage from, for example, a TE � 40 ms scan at 1.5T. Thisimage usually will have a resolution of no less than 0.5(read) � 1.0 (phase) � 2.0 mm (section). Despite the fact thatthe TE is so long, the image quality will be quite good apartfrom dephasing around the sinuses and in the air/tissue inter-face regions such as the pituitary gland. When compared withconventional gradient-echo scans with a TE of 25 ms (wherethe effects of veins are much less), SWI will tend to highlightsmall changes in susceptibility across a voxel as signal-inten-sity losses.

There are 2 key points to keep in mind when interpretingthe magnitude data. First, there is no large blooming effectbecause the resolution is so high. The higher the resolution,the less the blooming artifacts. Second, the parameters of thesequence are chosen so that the contrast is fairly flat betweennormal gray matter, white matter, and CSF, though a low-enough flip angle can keep the CSF brighter than the sur-rounding gray and white matter. Consequently, the magni-tude image clearly reveals areas with either short T2* or anoscillatory signal intensity in the magnitude image caused bythe presence of deoxyhemoglobin in the major veins. This lat-ter effect will cause a much larger signal-intensity loss than theactual T2* of the tissue because the signal intensity from thevein can cancel the signal intensity from the gray matter orwhite matter and, like water/fat cancellation, will cause a com-plete loss of signal intensity at the appropriate TE (this is why

the TE in SWI was chosen to be 40 ms in the first place). For avein of subvoxel size, this cancellation leads to a hypointensevoxel with respect to its surroundings. However, for a veincovering more than a few voxels, only in those voxels in whichboth vein and surrounding tissue are partial-volumed do wesee these signal-intensity cancellations and hypointense vox-els. A voxel containing only the intravenous signal intensitydoes not show this signal-intensity cancellation effect. Hence,the vein appears to have a bright signal intensity inside with ahypointense ring around it (Fig 7D). Nevertheless, followingphase-mask multiplication, the whole vein appears as a hy-pointense region in the processed SWI image (Fig 7F).

The second component to interpret in SWI data is thephase image. For a right-handed system, veins will look darkon the phase image (because the deoxygenated blood is para-magnetic relative to its surrounding tissue) and calcium willlook bright (because calcium is diamagnetic relative to thebrain tissue). Remnant phase artifacts may also appear by theair/tissue interfaces and above the sinuses, necessitating a care-ful recognition of the dark shadows in the phase images inthese areas as artifacts. Newer methods are making this less ofa problem.53,78,79 Interpreting a darkening in local areas inphase images is important for MS, where iron can be associ-ated with gray matter and white matter lesions. These iron-laden lesions will often correlate with the MS lesions seen onT2 or fluid-attenuated inversion recovery (FLAIR), but some-times they will not, presumably depending on the stage andtype of lesion.80 The phase difference can also be used to quan-

Fig 11. A sample mIP from 7T data, with a resolution of 215 � 215 � 1000 �, TE � 16ms, TR � 45 ms, FA � 25°, and mIP over 8 sections. Image courtesy of Ge Y and Barnes S.

Fig 12. A zoomed image from the same dataset shown in Fig 11 reveals a dark bandbetween the gray matter and white matter, which we assume represents the white matterarcuate fibers (diagonal arrows). The small vessels joining the pial veins are the venules(vertical arrows). Image courtesy of Ge Y and Barnes S.

AJNR Am J Neuroradiol 30:19 –30 � Jan 2009 � www.ajnr.org 9

tify iron or follow changes in iron deposition with time, tofollow the resolution of cerebral hemorrhages in patients withtrauma, or to monitor the development of microbleeds in pa-tients with dementia. In the future, phase will likely also beused to quantify oxygen saturation in major veins. Because theoriginal phase data have been filtered, we often refer to the newphase images as the SWI-filtered phase images.

The third component used in interpreting SWI is the finalprocessed SWI. These data have taken full advantage of T2*signal-intensity losses in the magnitude image and iron in-creases in the phase images to highlight both types of contrastin a single image. At this point, these images show many “blackholes,” most often representing the cross-sections of smallveins. How can we discriminate between these cross-sectionsand microbleeds? The best way to check for vessel connectivityis to take an mIP over a number of sections (most often 4).This is similar to using a maximum intensity projection (MIP)to see arteries in MR angiography (MRA). A point of interesthere is that the MIP of SWI data can be taken as well. Theresulting image, which is simply an SWI version of an MRA, isnot so impressive at 1.5T; but at 3T, this MRA begins to ap-proach the quality of a conventional MRA.53

A key point of practical interest is the orientation used forSWI. This is invariably transverse for a number of reasons. Theforemost reason is that the phase of veins perpendicular to themain field is, in theory, opposite to that of those parallel to themain field, with the latter showing a negative phase and theformer, a positive phase (equation 17). However, with an in-plane resolution to section-thickness aspect ratio of 2:4, thephase outside the small veins contributes more than the phaseinside, reversing the phase behavior so that all veins show anegative phase.30 This reversal of phase behavior makes theprocessing much easier because only a single negative phasemask needs to be used in the transverse orientation and thismakes the visualization of the veins in the SWI-processed im-ages much sharper.

Finally, we consider the contrast in the SWI data. Gener-ally, the low flip angles and intermediate TEs give SWI an oddmixture of contrast. The gray matter/white matter contrast isalmost nonexistent. Choosing the flip angle can play a role ingetting the CSF signal intensity close to the gray matter/whitematter signal intensity. This concept makes the magnitudeimages of little utility except as markers for microhemorrhagesor other susceptibility effects. We often choose a slightlyhigher flip angle that suppresses partially the CSF, making itslightly darker than gray matter or white matter and (similar toFLAIR) slightly T1-weighted. In that case, edema appearsbright, whereas the CSF remains suppressed. For patients withtumor, this has the effect of showing both the peritumoraledema as well as the bleeding inside the tumor.

Conversely, choosing a lower flip angle tends to highlightthe spin attenuation and makes CSF a bit brighter than graymatter or white matter to regain some contrast. The formerchoice of, for example, 20° at 1.5T is the most popular. Afterprocessing, the SWI takes advantage of the high iron contentin gray matter and begins to recreate gray matter/white mattercontrast. The phase image has a lot of resemblance to a T1-weighted image. This gets carried over to the SWI. At the end,to better delineate the nature of the vascular content, we per-form an MIP of the SWIs, usually over 4 adjacent images, tocreate an effective 8-mm-thick section. These mIP imagespresent a great tool for visualizing vessel connectivity and mi-crobleed locations with respect to the vasculature and otherstructures in the brain.

Future DirectionsTo date, SWI has had most of its clinical applications in theneuroimaging arena (this will be covered in Part 2 to be pub-lished in a later issue).81 Many of the ideas relating to ironquantification remain to be validated in animal models andfurther tested longitudinally in patients. Technical develop-ments such as creating susceptibility maps that remove allphase artifacts are a particularly exciting direction.78,82 Thiswould open the door to being able to use SWI in the spine, forexample. Other developing applications include imaging car-tilage, imaging calcium in atherosclerosis, imaging breast, andliver hemochromatosis.

AcknowledgmentsThe authors would like to thank Charbel Habib for manage-ment of all images and Alexander Boikov for reviewing themanuscript.

References1. Reichenbach JR, Venkatesan R, Schillinger DJ, et al. Small vessels in the human

brain: MR venography with deoxyhemoglobin as an intrinsic contrast agent.Radiology 1997;204:272–77

2. Baudendistel KT, Reichenbach JR, Metzner R, et al. Comparison of functionalMR-venography and EPI-BOLD fMRI at 1.5 T. Magn Reson Imaging1998;16:989 –91

3. Reichenbach JR, Essig M, Haacke EM, et al. High-resolution venography of thebrain using magnetic resonance imaging. MAGMA 1998;6:62– 69

4. Essig M, Reichenbach JR, Schad LR, et al. High-resolution MR venography ofcerebral arteriovenous malformations. Magn Reson Imaging 1999;17:1417–25

5. Lee BC, Kuppusamy K, Grueneich R, et al. Hemispheric language dominancein children demonstrated by functional magnetic resonance imaging. J ChildNeurol 1999;14:78 – 82

6. Lin W, Mukherjee P, An H, et al. Improving high-resolution MR bold veno-graphic imaging using a T1 reducing contrast agent. J Magn Reson Imaging1999;10:118 –23

7. Tan IL, van Schijndel RA, Pouwels PJ, et al. MR venography of multiple scle-rosis. AJNR Am J Neuroradiol 2000;21:1039 – 42

8. Reichenbach JR, Barth M, Haacke EM, et al. High-resolution MR venographyat 3.0 Tesla. J Comput Assist Tomogr 2000;24:949 –57

Table 2: Range of experimental parameters advised for SWI data collection at different field strengths*

FS FA (degrees) TR (ms) TE (ms) BW/pixel (Hz/pixel)1.5T 18–25 (20) 50–60 (60) 40 70–1003T 12–17 (12) 25–35 (30) 20 80–1004T 8–15 (12) 25–35 (28) 15 80–1207T 11–17 (14) 25–35 (30) 10–16 100–150

Note:—FS indicates field strength; FA, flip angle; BW, bandwidth.* The values given in parentheses for FA and TR are the ones generally used by the authors. TEs given here are optimized for maximum signal cancellation from venous vasculature. Slightlyshorter or longer TEs can be used for appropriate applications. However, note that for implementing shorter TEs, higher BW would be required, which would adversely affect the SNRof the image. One way to avoid this is to reduce the TR appropriately and cover more sections of the brain, which can recover some of the lost SNR.

10 Haacke � AJNR 30 � Jan 2009 � www.ajnr.org

9. Cheng YC, Haacke EM. Predicting BOLD signal changes as a function of bloodvolume fraction and resolution. NMR Biomed 2001;14:468 –77

10. Cheng YC, Haacke EM, Yu YJ. An exact form for the magnetic field density ofstates for a dipole. Magn Reson Imaging 2001;19:1017–23

11. Schad LR. Improved target volume characterization in stereotactic treatmentplanning of brain lesions by using high-resolution BOLD MR-venography.NMR Biomed 2001;14:478 – 83

12. Reichenbach JR, Haacke EM. High-resolution BOLD venographic imaging; awindow into brain function. NMR Biomed 2001;14:453– 67

13. Reichenbach JR, Jonetz-Mentzel L, Fitzek C, et al. High-resolution blood oxy-gen-level dependent MR venography (HRBV): a new technique. Neuroradiol-ogy 2001;43:364 – 69

14. Essig M, Reichenbach JR, Schad L, et al. High resolution MR-venography ofcerebral arteriovenous malformations [in German]. Radiologe 2001;41:288 –95

15. Haacke EM, Herigault G, Yu Y, et al. Observing tumor vascularity noninva-sively using magnetic resonance imaging. Image Analysis and Stereology2002;21:107–13

16. Tong KA, Ashwal S, Holshouser BA, et al. Hemorrhagic shearing lesions inchildren and adolescents with posttraumatic diffuse axonal injury: improveddetection and initial results. Radiology 2003;227:332–39

17. Essig M, Waschkies M, Wenz F, et al. Assessment of brain metastases withdynamic susceptibility-weighted contrast-enhanced MR imaging: initial re-sults. Radiology 2003;228:193–99

18. Abduljalil AM, Schmalbrock P, Novak V, et al. Enhanced gray and white mattercontrast of phase susceptibility-weighted images in ultra-high-field magneticresonance imaging. J Magn Reson Imaging 2003;18:284 –90

19. Hermier M, Nighoghossian N. Contribution of susceptibility-weighted imag-ing to acute stroke assessment. Stroke 2004;35:1989 –94

20. Wycliffe ND, Choe J, Holshouser B, et al. Reliability in detection of hemor-rhage in acute stroke by a new three-dimensional gradient recalled echo sus-ceptibility-weighted imaging technique compared to computed tomography:a retrospective study. J Magn Reson Imaging 2004;20:372–77

21. Tong KA, Ashwal S, Holshouser BA, et al. Diffuse axonal injury in children:clinical correlation with hemorrhagic lesions. Ann Neurol 2004;56:36 –50

22. Haacke EM, Xu Y, Cheng YC, et al. Susceptibility weighted imaging (SWI).Magn Reson Med 2004;52:612–18

23. Haddar D, Haacke E, Sehgal V, et al. Susceptibility weighted imaging: theoryand applications. J Radiol 2004;85:1901– 08

24. Rauscher A, Sedlacik J, Barth M, et al. Noninvasive assessment of vasculararchitecture and function during modulated blood oxygenation using sus-ceptibility weighted magnetic resonance imaging. Magn Reson Med2005;54:87–95

25. Sehgal V, Delproposto Z, Haacke EM, et al. Clinical applications of neuroim-aging with susceptibility-weighted imaging. J Magn Reson Imaging2005;22:439 –50

26. Haacke EM, Cheng NY, House MJ, et al. Imaging iron stores in the brain usingmagnetic resonance imaging. Magn Reson Imaging 2005;23:1–25

27. Mentzel HJ, Dieckmann A, Fitzek C, et al. Early diagnosis of cerebral involve-ment in Sturge-Weber syndrome using high-resolution BOLD MR venogra-phy. Pediatr Radiol 2005;35:85–90

28. Babikian T, Freier MC, Tong KA, et al. Susceptibility weighted imaging: neu-ropsychologic outcome and pediatric head injury. Pediatr Neurol2005;33:184 –94

29. Rauscher A, Sedlacik J, Barth M, et al. Magnetic susceptibility-weighted MRphase imaging of the human brain. AJNR Am J Neuroradiol 2005;26:736 – 42

30. Xu Y, Haacke EM. The role of voxel aspect ratio in determining apparentvascular phase behavior in susceptibility weighted imaging. Magn Reson Im-aging 2006;24:155– 60

31. Hamans BC, Barth M, Leenders WP, et al. Contrast enhanced susceptibilityweighted imaging (CE-SWI) of the mouse brain using ultrasmall superpara-magnetic iron oxide particles (USPIO). Z Med Phys 2006;16:269 –74

32. Sehgal V, Delproposto Z, Haddar D, et al. Susceptibility-weighted imaging tovisualize blood products and improve tumor contrast in the study of brainmasses. J Magn Reson Imaging 2006;24:41–51

33. Haacke EM. Susceptibility weighted imaging (SWI). Z Med Phys 2006;16:23734. Ashwal S, Babikian T, Gardner-Nichols J, et al. Susceptibility-weighted imag-

ing and proton magnetic resonance spectroscopy in assessment of outcomeafter pediatric traumatic brain injury. Arch Phys Med Rehabil 2006;87:S50 –58

35. Yoshida Y, Terae S, Kudo K, et al. Capillary telangiectasia of the brain stemdiagnosed by susceptibility-weighted imaging. J Comput Assist Tomogr2006;30:980 – 82

36. Noebauer-Huhmann IM, Pinker K, Barth M, et al. Contrast-enhanced, high-resolution, susceptibility-weighted magnetic resonance imaging of the brain:dose-dependent optimization at 3 Tesla and 1.5 Tesla in healthy volunteers.Invest Radiol 2006;41:249 –55

37. Haacke EM, DelProposto ZS, Chaturvedi S, et al. Imaging cerebral amyloidangiopathy with susceptibility-weighted imaging. AJNR Am J Neuroradiol2007;28:316 –17

38. Vinod Desai S, Bindu PS, Ravishankar S, et al. Relaxation and susceptibility

MRI characteristics in Hallervorden-Spatz syndrome. J Magn Reson Imaging2007;25:715–20

39. Akter M, Hirai T, Hiai Y, et al. Detection of hemorrhagic hypointense foci inthe brain on susceptibility-weighted imaging clinical and phantom studies.Acad Radiol 2007;14:1011–19

40. Larsen JP, Britt WI, Kido D, et al. Susceptibility weighted magnetic resonanceimaging in the evaluation of dementia. Radiology Case Reports 2007;2:102

41. Sedlacik J, Rauscher A, Reichenbach JR. Obtaining blood oxygenation levelsfrom MR signal behavior in the presence of single venous vessels. Magn ResonMed 2007;58:1035– 44

42. Pinker K, Noebauer-Huhmann IM, Stavrou I, et al. High-resolution contrast-enhanced, susceptibility-weighted MR imaging at 3T in patients with braintumors: correlation with positron-emission tomography and histopathologicfindings. AJNR Am J Neuroradiol 2007;28:1280 – 86

43. Shen Y, Kou Z, Kreipke CW, et al. In vivo measurement of tissue damage,oxygen saturation changes and blood flow changes after experimental trau-matic brain injury in rats using susceptibility weighted imaging. Magn ResonImaging 2007;25:219 –27

44. Ohta A, Naito K, Ohkubo M, et al. Study of susceptibility-weighted imaging(SWI) using a simple MR phantom [in Japanese]. Nippon Hoshasen GijutsuGakkai Zasshi 2007;63:1093–98

45. Thomas B, Somasundaram S, Thamburaj K, et al. Clinical applications of sus-ceptibility weighted MR imaging of the brain: a pictorial review. Neuroradiol-ogy 2008;50:105–16

46. de Souza JM, Domingues RC, Cruz LC, et al. Susceptibility-weighted imagingfor the evaluation of patients with familial cerebral cavernous malformations;a comparison with T2-weighted fast spin-echo and gradient-echo sequences.AJNR Am J Neuroradiol 2008;29:154 –58

47. Harder SL, Hopp KM, Ward H, et al. Mineralization of the deep gray matterwith age: a retrospective review with susceptibility-weighted MR imaging.AJNR Am J Neuroradiol 2008;29:176 – 83

48. Sedlacik J, Helm K, Rauscher A, et al. Investigations on the effect of caffeine oncerebral venous vessel contrast by using susceptibility-weighted imaging(SWI) at 1.5, 3 and 7 T. Neuroimage 2008;40:11–18

49. Somasundaram S, Kesavadas C, Thomas B. Susceptibility weighted imaging inholohemispheric venous angioma with cerebral hemiatrophy. Neurol India2008;56:104 – 05

50. Liu HL, Wai YY, Chen WS, et al. Hemorrhage detection during focused-ultra-sound induced blood-brain-barrier opening by using susceptibility-weightedmagnetic resonance imaging. Ultrasound Med Biol 2008;34:598 – 606

51. Koopmans PJ, Manniesing R, Niessen WJ, et al. MR venography of the humanbrain using susceptibility weighted imaging at very high field strength.MAGMA 2008;21:149 –58. Epub 2008 Jan 11

52. Fushimi Y, Miki Y, Togashi K, et al. A developmental venous anomaly present-ing atypical findings on susceptibility-weighted imaging. AJNR Am J Neuro-radiol 2008;29:E56

53. Du YP, Jin Z. Simultaneous acquisition of MR angiography and venography(MRAV). Magn Reson Med 2008;59:954 –58

54. Hammond KE, Lupo JM, Xu D, et al. Development of a robust method forgenerating 7.0 T multichannel phase images of the brain with application tonormal volunteers and patients with neurological diseases. Neuroimage2008;39:1682–92

55. Zhong K, Leupold J, von Elverfeldt D, et al. The molecular basis for gray andwhite matter contrast in phase imaging. Neuroimage 2008;40:1561– 66

56. Hu J, Yu Y, Juhasz C, et al. MR susceptibility weighted imaging (SWI) comple-ments conventional contrast enhanced T1 weighted MRI in characterizingbrain abnormalities of Sturge-Weber syndrome. J Magn Reson Imaging2008;28:300 – 07

57. Haacke EM, Lai S, Yablonskiy DA, et al. In-vivo validation of the BOLDmechanism: a review of signal changes in gradient-echo functional MRI in thepresence of flow. International Journal of Imaging Systems and Technology1995;6:153– 63

58. Haacke EM, Lai S, Reichenbach JR, et al. In vivo measurement of blood oxygensaturation using magnetic resonance imaging: a direct validation of the bloodoxygen level-dependent concept in functional brain imaging. Human BrainMapp 1997;5:341– 46

59. Akbudak E, Norberg RE, Conturo TE. Contrast-agent phase effects: an exper-imental system for analysis of susceptibility, concentration, and bolus inputfunction kinetics. Magn Reson Med 1997;38:990 –1002

60. Wang Y, Yu Y, Li D, et al. Artery and vein separation using susceptibility-dependent phase in contrast-enhanced MRA. J Magn Reson Imaging2000;12:661–70

61. Fernandez-Seara MA, Techawiboonwong A, Detre JA, et al. MR susceptometryfor measuring global brain oxygen extraction. Magn Reson Med2006;55:967–73

62. Haacke EM, Ayaz M, Khan A, et al. Establishing a baseline phase behavior inmagnetic resonance imaging to determine normal vs. abnormal iron contentin the brain. J Magn Reson Imaging 2007;26:256 – 64

63. Duyn JH, van Gelderen P, Li TQ, et al. High-field MRI of brain cortical sub-structure based on signal phase. Proc Natl Acad Sci U S A 2007;104:11796 – 801

64. Young IR, Bailes DR, Burl M, et al. Initial clinical evaluation of a whole body

AJNR Am J Neuroradiol 30:19 –30 � Jan 2009 � www.ajnr.org 11

nuclear magnetic resonance (NMR) tomograph. J Comput Assist Tomogr1982;6:1–18

65. Bydder GM, Steiner RE. NMR imaging of the brain. Neuroradiology1982;23:231– 40

66. Bailes DR, Young IR, Thomas DJ, et al. NMR imaging of the brain using spin-echo sequences. Clin Radiol 1982;33:395– 414

67. Edelstein WA, Hutchison JM, Johnson G, et al. Spin warp NMR imaging andapplications to human whole-body imaging. Phys Med Biol 1980;25:751–56

68. Haase A, Frahm J, Matthaei D, et al. Flash imaging: rapid NMR imaging usinglow flip-angle pulses. J Magn Reson 1986;67:258 – 66

69. Frahm J, Haase A, Matthaei D. Rapid 3-dimensional MR imaging using theflash technique. J Comput Assist Tomogr 1986;10:363– 68

70. Mugler JP 3rd, Brookeman JR. Three-dimensional magnetization-preparedrapid gradient-echo imaging (3D MP RAGE). Magn Reson Med 1990;15:152–57

71. Reichenbach JR, Venkatesan R, Yablonskiy DA, et al. Theory and application ofstatic field inhomogeneity effects in gradient-echo imaging. J Magn Reson Im-aging 1997;7:266 –79

72. Haacke EM, Brown R, Thompson M, et al. Magnetic Resonance Imaging: PhysicalPrinciples and Sequence Design. New York: Wiley; 1999:5

73. Schenck JF. The role of magnetic susceptibility in magnetic resonanceimaging: MRI magnetic compatibility of the first and second kinds. Med Phys1996;23:815–50

74. Saini S, Frankel RB, Stark DD, et al. Magnetism: a primer and review. AJR Am JRoentgenol 1988;150:735– 43

75. Chu SC, Xu Y, Balschi JA, et al. Bulk magnetic susceptibility shifts in NMR

studies of compartmentalized samples: use of paramagnetic reagents. MagnReson Med 1990;13:239 – 62

76. Weisskoff RM, Kiihne S. MRI susceptometry: image-based measurement ofabsolute susceptibility of MR contrast agents and human blood. Magn ResonMed 1992;24:375– 83

77. Spees WM, Yablonskiy DA, Oswood MC, et al. Water proton MR properties ofhuman blood at 1.5 Tesla: magnetic susceptibility, T(1), T(2), T*(2), and non-Lorentzian signal behavior. Magn Reson Med 2001;45:533– 42

78. Neelavalli J, Cheng YC, Haacke EM. Removal of Air/Tissue Interface FieldEffects in Susceptibility Weighted Imaging. Proceedings of International So-ciety of Magnetic Resonance in Medicine. Toronto, Ontario, Canada; 3–9 May2008. Abstract 3499

79. Rauscher A, Barth M, Reichenbach JR, et al. Automated unwrapping of MRphase images applied to BOLD MR-venography at 3 Tesla. J Magn Reson Im-aging 2003;18:175– 80

80. Haacke EM, Makki M, Ge Y, et al. Characterizing iron deposition in multiplesclerosis lesions using susceptibility weighted imaging. J Magn Reson Imaging2009 (In press).

81. Tong KA, Ashwal S, Obenaus A, et al. Susceptibility-weighted MR imaging: areview of clinical applications in children. AJNR Am J Neuroradiol2008;29:9 –17

82. de Rochefort L, Brown R, Prince MR, Wang Y. Quantitative MR susceptibilitymapping using piece-wise constant regularized inversion of the magneticfield. Magn Reson Med 2008;60:1003– 09

12 Haacke � AJNR 30 � Jan 2009 � www.ajnr.org